Case study Finance automation · Higher education · Singapore

A university's month-end close went from three days to under ten minutes.

Faster closes, fewer errors: how we rebuilt month-end reconciliation for a globally distributed finance team.

Our client, the incubation arm of a leading Singaporean public university, reconciled revenue, expenses, and budget for teams spread across the world. The reconciliation and verification took about three days each month.

We turned three days of spreadsheet work into an automated run that finishes in under ten minutes, and gave each team visibility into their numbers.

Results Month-end close · four closes in production
A monthly close
10 min
Consolidated, reconciled and published — under ten minutes, down from three days
In production
4 closes
March to June 2026, no issues reported
Scoped dashboards
5
Live, and each team sees only its own numbers
Full rebuild
6 weeks
Across P&L, receivables and payables reconciliation
The challenge

A mistake could travel a long way before anyone caught it.

Every month, our client's finance office rebuilt the university's financial picture from two files: a full SAP export of revenue line items and an HR file of expenses. The process was repetitive and data-intensive: copy the rows, reformat codes, reconcile totals, check the month against the last, and only then send the numbers to the teams that ran on them.

The process had limited validation and only a human verification safety net. One team drove it end to end, and payables ran on a separate manual track: figures managed on a spreadsheet. These numbers fed budgets and decisions for teams around the world, so a mistake could travel a long way before correction.

Our approach

Automate the mechanical work. Keep human judgment on the numbers.

realfast builds AI-first, while keeping human judgment on numbers. We automate mechanical work, the parsing, transforming and reconciling, and give the finance team charge of insights and decisions. The workbook held years of the client's edits, so everything we built would follow existing patterns. All reconciliations had to prove their own totals, so every run checks itself.

How we delivered

Upload, process, reconcile, export — and nothing publishes until the totals match.

We built the month close as a pipeline: upload, process, reconcile, export. The engine takes full monthly Revenue and Cost source files, and processes each month on its own, adding only what is new and leaving every prior month untouched. Before anything is published, it checks that its totals match the original sources.

< 10 min a close

down from about three days of copying, reformatting and reconciling, every month.

Before — one close, manual
~3 days
After — one close, automated
< 10 min

P&L consolidation, reconciliation and publishing for one monthly close. Before, about three days of manual work; after, under ten minutes — verified across the March–June 2026 closes, run in production.

Month-end close

One pipeline, and a gate it has to pass

Each month processed on its own: only what is new is added, every prior month left untouched.

Upload Full monthly Revenue & Cost source files
Process Company-specific logic, one month at a time
Reconcile · gate Totals must match the source, or nothing publishes
Export The consolidated financial report
Five scoped dashboards P&L, Budget-vs-Actuals, Cost Details, Receivables and Payables, each scoped to team access: every team sees only its own numbers.
The gate is the point. Upload, process and export are speed; the reconciliation check is why the finance office was willing to hand over a live close.
MONTH-END CLOSE AI-First Finance Reconciliation INPUTS Revenue & Cost Systems (SAP / HR / etc.) PROCESS company-specific logic to pivot data RECONCILE · GATE totals must match before anything publishes PASS → ship    FAIL → stop OUTPUT Financial Report CLOSE TIME under 10 min FIVE SCOPED DASHBOARDS P&L Budget-vs-Actuals Cost Details Receivables Payables Row-level security: each team sees only its own numbers.
the pipeline as drawn: realfast · anonymised

We then built a validated one-click upload and reconciliation for their Accounts Payable, in under two weeks. Accounts Receivable got the same treatment, with aging computed automatically. Work that used to take time to prepare now arrives ready to use, at the click of a button.

Finally, we built real-time views for visibility. We piped the reconciled data into five live dashboards, P&L, Budget-vs-Actuals, Cost Details, Receivables, and Payables, each scoped to team access. Correcting the budget and variance logic along the way surfaced planning and spend gaps the spreadsheets had hidden.

The pipeline has run every monthly close since March 2026: four closes, no issues reported.

What we learned
Trust is built by enforcing strict verification and validation.

Our client handed over a live close because the tool guarded their historical edits and proved its totals every time.

Fix the logic and the questions change.

Reducing time taken to reconcile reports from global offices freed up the finance team's focus and time to dig into insights for decision-making in real-time.

We automated the drudgery, not the judgment.

Partnership

A precise workflow, and reverse-demos of their own closes.

Our client's finance office had a precise, well-documented workflow for each month's close. It was just time-consuming and vulnerable to human error, given the scale and repetitiveness of the reconciliation. Their reverse-demos of previous closes, walking us through their logic, preferences, and pain points, made it far easier to zoom in and deliver on a critical application quickly.

We ran an almost-daily delivery cycle: testing with users, collecting feedback, and finding bugs in the first half of the day; fixing them and shipping a revamped experience by the next morning. That compressed the discovery-to-delivery loop.

Talk to us

Still closing the month by hand?

If a repetitive, high-stakes workflow eats days of your team's month, that time is recoverable. We can prove it on your own files, in weeks. That's where realfast starts. Talk to us.

Let's go

Thanks, we're thrilled you're here. We'll be in touch shortly to set up your session.

Something went wrong

We couldn't submit your request. Please try again, or email us at hello@realfast.ai.

Anonymised · the incubation arm of a leading Singaporean public university (its finance office). Figures as delivered and verified in production across the March–June 2026 closes. No client figures published.

Take this case study with you

Save a clean, print-ready PDF to share internally or send to your team.